Effects of emotional agents on human players in the public goods game

Abstract

In previous work we chose the public goods game as a small and adequate example scenario for emotional agents. In our experiment we focus on the influence of emotional reactions of the emotional agents on a human player. The artificial emotions of the agents are generated by a model of emotion which is adapted to the scenario. In our experiment we compare the reactions of the human player in a scenario with and without visual feedback of the other agents. As a visual feedback we use the ECA Greta to embody the emotions generated by our model. Adequate emotion recognition could be proven by an online survey validating the correct valence and intensity perception of the relevant emotions. As one of the first results, a significantly higher amount of investment can be stated for the human player in case he or she sees the facial representation of the emotional state of the other (virtual) players during the public goods game. After splitting up the game in two sections, it could be shown that this significant difference is only true for the second part of the game.